本研究基于KISS(keep it simple and stupid)算法,利用似然比测试直接为矩阵模式定义度量,解决了现有大多数度量学习算法需要经过复杂优化过程的问题。通过在似然比测试中有目的地引入矩阵正态分布,该度量无需将矩阵模式通过向量化的方...本研究基于KISS(keep it simple and stupid)算法,利用似然比测试直接为矩阵模式定义度量,解决了现有大多数度量学习算法需要经过复杂优化过程的问题。通过在似然比测试中有目的地引入矩阵正态分布,该度量无需将矩阵模式通过向量化的方法变成向量模式,因而具有如下优点:(1)能够避免维数灾难;(2)比KISS更鲁棒;(3)无需计算大矩阵的逆和特征值分解,因此计算远快于KISS算法。最终的实验验证了该算法的优势。展开更多
Trend test in dose-response has been a central problem in medicine. This paper treats the problem of comparing umbrella pattern treatment effects. Under an ordered m × r × k table, this article considers tes...Trend test in dose-response has been a central problem in medicine. This paper treats the problem of comparing umbrella pattern treatment effects. Under an ordered m × r × k table, this article considers testing the hypothesis that all multinomial populations are conditional independence against the alternative that they are in an umbrella trend. For this hypothesis test problem, this article introduces a model-free test method by using likelihood ratio test statistic and gives the asymptotic distribution of the test statistic. Simulation study is conducted to compare the empirical power per- formed via the proposed method and others. Finally, two real data are studied to illustrate the validity of the proposed method.展开更多
This paper proposes the corrected likelihood ratio test （LRT） and large-dimensional trace criterion to test the independence of two large sets of multivariate variables of dimensions P1 and P2 when the dimensions P ...This paper proposes the corrected likelihood ratio test （LRT） and large-dimensional trace criterion to test the independence of two large sets of multivariate variables of dimensions P1 and P2 when the dimensions P = P1 ＋ P2 and the sample size n tend to infinity simultaneously and proportionally. Both theoretical and simulation results demonstrate that the traditional X2 approximation of the LRT performs poorly when the dimension p is large relative to the sample size n, while the corrected LRT and large-dimensional trace criterion behave well when the dimension is either small or large relative to the sample size. Moreover, the trace criterion can be used in the case of p 〉 n, while the corrected LRT is unfeasible due to the loss of definition.展开更多
文摘本研究基于KISS(keep it simple and stupid)算法,利用似然比测试直接为矩阵模式定义度量,解决了现有大多数度量学习算法需要经过复杂优化过程的问题。通过在似然比测试中有目的地引入矩阵正态分布,该度量无需将矩阵模式通过向量化的方法变成向量模式,因而具有如下优点:(1)能够避免维数灾难;(2)比KISS更鲁棒;(3)无需计算大矩阵的逆和特征值分解,因此计算远快于KISS算法。最终的实验验证了该算法的优势。
基金supported by the National Natural Science Foundation of China under Grant No.10771163
文摘Trend test in dose-response has been a central problem in medicine. This paper treats the problem of comparing umbrella pattern treatment effects. Under an ordered m × r × k table, this article considers testing the hypothesis that all multinomial populations are conditional independence against the alternative that they are in an umbrella trend. For this hypothesis test problem, this article introduces a model-free test method by using likelihood ratio test statistic and gives the asymptotic distribution of the test statistic. Simulation study is conducted to compare the empirical power per- formed via the proposed method and others. Finally, two real data are studied to illustrate the validity of the proposed method.
基金supported by National Natural Science Foundation of China(Grant Nos.11101181,11171057,11171058 and 11071035)Research Fund for the Doctoral Program of Higher Education of China(Grant No.20110061120005)+1 种基金NECT-11-0616,PCSIRTthe Fundamental Research Funds for the Central Universities
文摘This paper proposes the corrected likelihood ratio test （LRT） and large-dimensional trace criterion to test the independence of two large sets of multivariate variables of dimensions P1 and P2 when the dimensions P = P1 ＋ P2 and the sample size n tend to infinity simultaneously and proportionally. Both theoretical and simulation results demonstrate that the traditional X2 approximation of the LRT performs poorly when the dimension p is large relative to the sample size n, while the corrected LRT and large-dimensional trace criterion behave well when the dimension is either small or large relative to the sample size. Moreover, the trace criterion can be used in the case of p 〉 n, while the corrected LRT is unfeasible due to the loss of definition.
